Implement a computer vision application that re-evaluates recorded basketball games (e.g. Olympic finals) in the following way: (1) Employ skeleton tracking (e.g. with OpenPose) on the game in order to (a) judge the height of the players (relativ to the height of the baskets is sufficient), (b) record their scoring; (2) based on the height analysis, compute how easy it is for the players to score (= the taller the easier) by doing literature research and, if necessary, a brief empirical evaluation; (3) based on these data, re-compute the scores of at least three games by using the computed easiness factor as a weight on the scores of the players.
Requires excellent knowledge in data science on the sub-symbolic level, in particular computer vision.